Our keyserver has been operational for about 4 months now. I noticed that disk usage had almost doubled. It was about 6G when we started and close to 12G after 4 months. I decided to create a new dump and rebuild the database. That brought it back down to about 6G. Here's a brief outline of the procedure:
* Note the current number of keys your server has. This will be a good indication of if you get everything working again. * Clean the database with "sks cleandb" * Create a new dump ** Create the directory, e.g. dump.new ** Generate the dump with "sks dump 15000 dump.new" * Shut down the server. * Remove or rename the KDB and PTree directories. * Rename the dump.new directory to dump. * Run sks_build.sh. This will create new KDB and PTree directories. * Start the keyserver and verify that you have about the same number of keys you started with. Our server is a virtual machine with 2 2.6GHz CPUs and 4GB memory. The dump took about 30 min, fastbuild 11 min, clean 2 min, pbuild 30 min. (The last three are part of sks_build.sh.) Since I did not shut down the server before the dump, there is a small chance that someone sent us some keys that did not get forwarded to another keyserver. The risk is probably pretty small, but you could shutdown before the dump if you're worried about it. I couldn't really find this all in one place, so I thought I'd post it here to get in the archive. Does the procedure seem reasonable? Phil _______________________________________________ Sks-devel mailing list Sks-devel@nongnu.org https://lists.nongnu.org/mailman/listinfo/sks-devel
